Filter
Top Products
Traffic engineering
186 Avaya Application Solutions IP Telephony Deployment Guide
Example 2: Uniform Distribution model
In the case of a stand-alone Avaya system, the Uniform Distribution model works on the
assumption that when a given station places an intercom call, the call is equally likely to
terminate at any of the other stations in the entire system. Analogous statements regarding this
model can also be made for inbound trunk calls and outbound trunk calls. Specifically, any
inbound call is equally likely to terminate at any of the stations in the system, and any outbound
call is equally likely to have been originated by any of the stations in the system. The
fundamental concept underlying the Uniform Distribution model is that stations are essentially
indistinguishable from one another from a traffic engineering point of view. This model is usually
the most appropriate option when engineering a system for which little or no information about
the nature of the various stations exists. This model will now be applied to the system that is
described in Example 1: Station usage
.
The design criteria for Example 1: Station usage
was one-third of all calls being intercom,
one-third being inbound PSTN, and one-third being outbound PSTN. From the station usages
that are listed in Example 1: Station usage
, it follows that the total station usage in Atlanta is 195
Erlangs, the total in Boston is 72 Erlangs, and the total in Cleveland is 34 Erlangs, for a
system-wide total of 301 Erlangs of station usage. Under the “one-third intercom, one-third
inbound, one-third outbound” assumption, this corresponds to a system-wide total of 75 Erlangs
of intercom call usage, 75 Erlangs of inbound call usage, and 75 Erlangs of outbound call usage
(rounding to the nearest Erlang in each case). To verify this, first consider the fact that all three
components are equal (each is 75 Erlangs) satisfies the “one-third, one-third, one-third”
requirement. Furthermore, since 75 Erlangs of intercom call usage corresponds to 150 Erlangs
of station usage, 75 Erlangs of inbound call usage corresponds to 75 Erlangs of station usage,
and 75 Erlangs of outbound call usage corresponds to 75 Erlangs of station usage, there is a
total of 150 + 75 + 75 = 300 Erlangs of station usage. This agrees with the specified 301 Erlangs
if one ignores error due to rounding off.
One could assume in this example that each PSTN trunk is capable of carrying both inbound
calls and outbound calls. Trunks are normally engineered to a desired Grade of Service (GOS),
or blocking level. A commonly used GOS for trunks is P01, which represents a nominal blocking
rate of 1 out of every 100 call attempts. To determine how many trunks are needed to attain
P01, one must know the call traffic load to be carried by those trunks. Both inbound call usage
and outbound call usage are included in that load.
Note:
Note: If IP Softphone telecommuters were used in this example, they would have also
contributed toward trunk load. Although the signaling link between a
telecommuter and the Communication Manager system to which the
telecommuter is registered is carried over IP, the media flow between the two
uses a PSTN trunk.
Example 1: Station usage
indicates that the total load to be carried by the trunks is 75 + 75 =
150 Erlangs, which accounts for both inbound and outbound PSTN call usage. Use of the
standard Erlang blocking model indicates that 171 trunks (DS0s) would be required to carry the
150 Erlangs of trunk call usage at P01. However, one must consider the trunk selection process
for PSTN calls.